0
\$\begingroup\$

TI recommends their low-data rate transceivers to cope with long stub lengths for RS485 networks -- eg Page 3 of http://www.ti.com/lit/an/slla272c/slla272c.pdf

I want to use an isolated interface, so I was thinking about using the il3085 chip, but it has fast rise/fall times for high speeds (I'm using 115kbaud at most, so this is unwanted overkill).

Can I add some series resistance to the il3085 and be OK (how do you calculate that?), or should I switch to the sn65hvd1780 and give up on isolation?

\$\endgroup\$
  • \$\begingroup\$ What stub length do you need>? A filter can be designed to increase stub lengths for 115kbd at some increase in ISI or edge jitter but reduced return loss from open stub. \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Nov 15 '17 at 20:17
  • \$\begingroup\$ Why not just pair a low-speed transceiver with a separate isolator? \$\endgroup\$ – Abe Karplus Nov 15 '17 at 20:31
  • \$\begingroup\$ Ideally around 4 meters/12 feet (the wire that usually comes on it is 3m long). I was figuring I could add a resistor to the A&B lines (and maybe a capacitor on the network side after that), but I don't remember how to calculate the appropriate values... I liked the il3085 because it's compact (only good for 300V, but that should be plenty) and it has its own power supply (as opposed the the TI ones where you need to build an isolated 5V power supply), but I'm open to suggestions. \$\endgroup\$ – Thor Johnson Nov 15 '17 at 20:51
  • \$\begingroup\$ @ThorJohnson Not sure where you're getting that the IL3085 has its own power supply: the datasheet suggests you need to provide isolated power. \$\endgroup\$ – Abe Karplus Nov 15 '17 at 22:29
  • \$\begingroup\$ @AbeKarplus Ah crap. Nevermind. I saw that the EVB just had capacitors on it and ran off... looking back on it, it says "hook to 5V power supply", so it's more complicated than I wanted it to be. \$\endgroup\$ – Thor Johnson Nov 16 '17 at 16:31
1
\$\begingroup\$

The rise time can be reduced to increase stub length, as long as satisfies some challenges;

  • =>10x the stub prop. delay
  • =<10% of a data bit
  • Zo error =<25%

    • An open stub reflects 100% of the signal with twice the length delay at bus
    • An output cap in LPF causes Zo mismatch on attenuation with variable reflection as Zc(f) drops with rising hamonics.
    • steep filters , even Gaussian, can cause ISI degradation, unless "raised Cosine" type which overshoot with zero crossings same as data.
\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.